Risk segment selection: how to find segments with the best loss ratios

Data-driven segment selection enables insurers to identify attractive risk segments outside of their claims experience, so they can optimise their portfolio mix to grow premium at equivalent or better loss ratios.

Targeting specific risk segments has been an integral part of business strategy for insurers since 1667 when economist Nicholas Barbon established the first ‘fire only’ insurance company which focused on insuring brick and frame homes after the great fire of London.

CURRENT SITUATION

Segment selection plays a vital role in driving portfolio level profitability and growth but comes with a major challenge: if a segment is only attractive retrospectively once claims can be compared to premium, how does one find the best segments in the market without writing the whole market? 

Segment decisions today are based on a mixture of claims data and underwriting intuition. They are also heavily influenced by existing distribution networks. Both factors constrain insurer's ability to identify and access new and profitable classes of risk.

The lens of this article will focus on segment selection from a profitability standpoint, auxiliary issues relating to premium growth and distribution will be addressed at a later point. 

THE PROBLEM

Operating within their limited portfolio experience, insurers use a combination of historical claims and underwriting intuition to make decisions about which risk segments to focus on in the future. Acute diversification and reliance on history means that insurers operate with a backward-looking, skewed view of the market, making it difficult for them to identify segment opportunities that can improve their loss ratio. 

Two discrete problems exist. The first is identifying - in spite of limited and often statistically insignificant data - which segments out of their current appetite yield the most attractive loss ratio and premium growth opportunities. The second, a related but more valuable problem, is building a menu of all possible segments available in the market, many of which lie outside of underwriting experience and exposure to return the best loss ratio. 

PATH DEPENDANCY

Segment choices exert path dependency on the future profitability of the insurer’s portfolio. To decrease volatility and amass claims experience, insurers will converge towards what is known and backed up with limited experiential data rather than moving into the unknown where loss ratios are more difficult to predict.  

The path dependency of portfolio choices means that the probability of further concentration in known segments increases with each claim and unit of exposure. The outcome is a tendency towards binary decisions: wholesale persistence to accumulate experience, or divestment from a line entirely as the ability to dynamically tune or revise the mix between segments is impaired. 

The sudden switch between underwriting experience accumulation reversing into a wholesale exit can be seen in the tow truck segment during 2015 and 2016. This is a recent, indicative example where nine US carriers exited the segment triggered by Progressive’s decision to pull their coverage nationwide in light of rising losses. 

Therefore the segment in which an insurer starts out is the defining influence on the future profitability of their portfolio. As illustrated in Figure 1, insurers typically converge towards local maxima of profitability - the best segment within a stone’s throw of where they started, rather than the more desirable global maxima of profitability.

The combination of incomplete, variegated experience and path dependency dynamics creates vicious - not virtuous - cycles.

Experience based iterative underwriting appetite adjustments lead to path dependency, meaning that insurers miss or never discover the best segments. 

Path Dependancy

SOLUTION

Identifying the most attractive segments in a market requires an exhaustive view of the entire insurable population at the resolution of each insurable risk. 

A comparison between population performance vs. portfolio performance enables a cycle of continual optimisation - for example, comparing loss ratios of chosen segments to those outside of appetite to drive continuous revision of the portfolio mix. This has been the practice in asset management for almost a decade, where a market view (from Bloomberg) is compared to a portfolio, with appetite and selection criteria adjusted continuously.

Insurance today is the opposite and can be characterised as a configuration of many fractured, partially overlapping stock markets reflecting the thin towers of experience built up by insurers. 

The Cytora Risk Engine provides Chief Underwriting Officers and Chief Actuaries with a population scale view of risk, enabling them to continuously benchmark and optimise the performance of their selected segments against the performance of the entire insurable population.

To learn more about how Cytora can help you to optimise your portfolio and grow premium, please contact us.